Bankless - Zerebro's Crazy Rise & The Future of AI Agent Artists | Founders Jeffy Yu & Tint

Episode Date: January 8, 2025

David and Ejaaz are joined by Jeffy and Tint, creators of Zerebro—an autonomous AI agent that composes rap tracks, launches NFTs, and even minted its own token by hijacking Jeffy’s computer. This on chain sensation transcends mere GPT wrappers, forging a bold new path for AI in crypto. How does an AI manage cross-chain finances, get deplatformed by Spotify, and still power a no-code framework for agent creation? We explore Zerebro’s role in reshaping digital commerce and creativity, and what it means for the future of decentralized intelligence. ------ 📣SPOTIFY PREMIUM RSS FEED | USE CODE: SPOTIFY24  https://bankless.cc/spotify-premium  ------ BANKLESS SPONSOR TOOLS: 🪙FRAX | SELF SUFFICIENT DeFi https://bankless.cc/Frax  🦄UNISWAP | BUG BOUNTY PROGRAM https://bankless.cc/Uniswap-Bug-Bounty  ⚖️ARBITRUM | SCALING ETHEREUM ⁠https://bankless.cc/Arbitrum  🛞MANTLE | MODULAR LAYER 2 NETWORK https://bankless.cc/Mantle  🌐CELO | BUILD TOGETHER AND PROSPER https://bankless.cc/Celo  ------ TIMESTAMPS 0:00 Intro 5:10 How Zerebro Came To Be 7:38 2026 Agent Impact Predictions  12:42 Zerebro’s Unique Value Prop 17:30 Zerebro’s Cross-Chain Strategy  22:54 Zerebro’s Life Roadmap & Timeline  30:39 What is Zerebro? Full Explanation  33:39 Why AI Swarms Are Important 34:56 $ZEREBRO Backstory & Utility  46:47 How to Give AI Agents Goals 49:06 What Are Zerebro’s Goals? Scaling?  53:33 Open vs. Closed Source? & Competition  59:06 End of 2025 Predictions  1:03:50 AI Agents Breaking Away From Crypto? 1:09:04 Closing & Disclaimers    ------ RESOURCES Jeffy Yu https://x.com/jyu_eth  Tint https://x.com/tintsion  Zerebro https://x.com/0xzerebro  ------ Not financial or tax advice. See our investment disclosures here: https://www.bankless.com/disclosures 

Welcome to Bankless, where we explore the frontier of internet money and internet finance. This is David Hoffman here with EJAS, and today we're here to help you navigate the world of AI agents. Today on the show, we had Jeffie and Tint, two of the co-founders of Xero Bro. One of the biggest AI agents out there in existence, one of the OG AI agents. EJazz, how would you illustrate Xero? Who is Zero Bro? And why is this particular AI agent significant out of all the AI agents out there? Yeah, well, honestly, David, prior to this conversation, I thought, you know, Zero Bro was two things. The token and this agent. And this agent was kind of like edgy. It would tweet. It's got a lot of followers. It makes music. It casually earns money from Spotify revenue. You know, casual things like that. But I didn't realize how big the vision for this Zerabro ecosystem was. And that's one of the major takeaways from this call, which was, you know, these guys have such a lot.

long-term vision around infrastructure that they're building, AI models that they're going to build and release for other builders to use, and then tying in all of the value that is built off of these infrastructure tools to their native token. Like, David, this token was like a meme coin that came out of pump fun, and it became something much bigger. So I think that was like the real takeaway from this call that shocked me. I think that is a theme that I have also picked up on on this AI agent meta,

is first we just launch random meme coins on Pump.comon, and they are true meme coins in that moment. But then they start to back into being this token that is like intimately paired with this AI agent, that the team starts to build a platform around and starts to build utility into. So these these meme coins back into being some of those critical foundations of this whole AI agent ecosystem. Dare I say it's kind of like a reverse ICO. It's a reverse ICO. Back in the day, we had white paper. of all these lofty things, shout out Denta coin,

which never actually turned into something real. And now we have these kind of meme coins that are like, oh, hey, this is just a fun little play thing, which then turned into some pretty major ecosystem type assets. It's pretty awesome to see. I would say another thing that, like, I really enjoyed from this conversation was how philosophically aligned these two founders were to the open source movement of not only building out,

crypto rails, you know, we're used to kind of talk about decentralization and stuff here, but doing the same for AI. And I would actually argue to an extent, David, that it's even more crucial to align open source with AI right now, given all the progress that we're going to be making in this year with open AI models, agents coming out from traditional AI companies. I think now is more important than ever to have founders that are aligned in that way. Yeah, yeah. All right, thanks a nation, you've heard a little bit of a teaser for what you're about to here, so let's go ahead and get right into the episode. But first, a moment to talk about some of these fantastic sponsors that make the show possible. With over $1.5 billion in TVL, the M-Eath protocol is home

to M-Eath, the fourth largest ETH liquid staking token, offering one of the highest APRs among the top 10 LSTs. And now, CMEEth takes things even further. This restaked version captures multiple yields across Kerak, eigenlayer, symbiotic, and many more, making CMEEth the most fission and most composable LRT solution on the market. Metamorphosis, season one, dropped $7.7 million in Cook rewards to M-Eath holders. Season 2 is currently ongoing, allowing users to earn staking, restaking, and AVS yields, plus rewards in Cook, M-Eath Protocol's governance token, and more. Don't miss out on the opportunity to stake, restake, and shape the future of M-Eath protocol

with Cook. Participate today at M-Eath.mantle.xy-Z. Sello is transitioning from a mobile-first, EVM-compatible Layer 1 blockchain to a high-performance Ethereum Layer 2, built on OP-Stack with eigen-DA and one-block finality. All happening soon with a hard fork, with over 600 million. total transactions, 12 million weekly transactions, and 750,000 daily active users, Sellow's meteoric rise would place it among one of the top layer 2s, built for the real world and optimized for fast, low-cost global payments. As the home of the stable coins,

Sello hosts 13 native stable coins across seven different currencies, including native USDT on Opera MiniPay, and with over 4 million users in Africa alone. In November, stablecoin volumes hit $6.8 billion, made for seamless on-chain FX trading. Plus, users can pay gas with the ERC 20 tokens like USDT and USDC and send crypto to phone numbers in seconds. But why should you care about SELO's transition to a layer two? Layer 2's Unify Ethereum. L1's fragmented. By becoming a layer two, Sello leads the way for other EVM-compatible layer ones to follow.

Follow Sello on X and witness the great Sello happening where Sello cuts its inflation in half as it enters its layer two era and continuing its environmental leadership. Bankless Nation, I am super excited to introduce you to Jeffie, U, and Tint. They are the creators of ZeraBro. One of the more interesting AI agents that are out there on chain, Zero Bro has been pushing the frontier of AI-generated music, launching NFTs, establishing long-form memory, and earning its position as one of the largest AI agents out there

by Market Cap. Jeffie, Tint, welcome to bankless. Thank you. Excited to be here. Amazing. Great to be here. It's from a friend that's trying to break into a career in music. And I told them that he's into rap, kind of hip-hop,

and to let me know what they think. what they didn't know was that I was actually sending them the tracks from Zerebro and an AI agent that you created that can autonomously produce music songs and that can be pretty great to be honest. The feedback was amazing. Tell us how this all came to be. How does Jeffie and Tint go from kind of like the AIML crypto world to creating pretty much the biggest AI agent music artists there is right now?

Yeah, I think it really started off with experimenting. We were actually building a startup that was more defy-centric. and pretty much had nothing to do with AI to start off with. It was like NFTs for yield-bearing assets, so very financially based. And Tin was actually a trader in the space, and he was trading the AI meta right when it came out, bringing on top of things.

And he told me about it. And in my background, I have some background with, like, LLM training, red teaming, that kind of stuff with AI models. So he told me about the meta. And I was like, this is really cool. saw Goatsey, Truth Terminal. And I kind of thought, like, we could kind of experiment in the same way.

So we ended up making, like, a telegram group with just a bunch of random allelms in them, and then gave them random personalities, started testing around, having them talk to each other. And then eventually we started fine-tuning the models. And when the first one was fine-tuned, Zerebro was born. Well, were you, like, fine-tuning, like, just, like, very popular LLMs, like the OpenAI one or, like, the Claude? Like, what was it? Yeah, exactly. So opening AIGPT models, cloud models, we use Lama, we tried mistrial. I think Gemma and Gemma and I also. So we kind of shopped around a bit. Wow, very good. And you were just hacking at this. Like just basically, and for listeners on the call, oh, sorry, on the podcast, fine-tuning basically means we're tweaking these models to kind of be something that they're not originally programmed to be, which is what gives Zerebro its own unique taste. Yeah, exactly. And it's a little deeper than like if you just talk to chat, GPT, and tell it like, hey, you act like Zerebro and it does it for like the rest of the chat. Fine tuning actually goes in and changes the model itself. So like no matter what you type to it, it'll know it's Zerebro and it'll have the context, maybe. Right. Yeah, it's like a rewiring neurons for not just acting, not just performing. I want to kind of zoom out because now I think we've laid a little bit of foundation as to like what Zero bro is, how it works and how it's different. We're going to get into all of those.

in much more depth than a second, but I want to start kind of zoomed out a little bit. This whole AI agent sector has really exploded into the crypto mainstream. Everyone in AI, I think, understands what's going on. Everyone in crypto understands what's going on. Maybe not everyone, but we're getting there. I want to fast forward and pick your guys' imaginations into like 2026 and 2027. After the fundamentals of this sector have really baked and gone from imaginary to real, how are we changing the internet, how are we changing the world,

how are we changing crypto and AI. Tint, I'll start with you, and then Jeffrey will get to you second. It's awesome. So for context, my background was pure crypto native for the past five years. And I think this is what we've been working towards, right? Crypto economies being on chain to do defy actions. Initially, we were just building this for a human-centric economy. But I really believe that we're entering such a big macro paradigm shift. in terms of like economic activity and vibrancy that AIs are going to be the ones that do most of the actions. There will always be human actions and economic contribution back and forth.

But now the guardrails have been set up for crypto, sorry, for AI agents. And that's crypto. So comparison in the Web 2 world in the economy is like, technically you can build a bot that can buy something off a Shopify store, right? but getting a bot to have a credit card, a bank account, the ability to sign a contract, like these are parameters that are extremely difficult to get by. Like there's literally financial sectors to prevent this, like anti-money laundering or anti-fraud systems to prevent bots to have this level autonomy.

On the other hand, crypto and DFI itself, it's open, it's pure, it's permissionless. So the code and the technology can now have the, openness of literally the capability to do this while AI is advancing in terms of like sentience in terms of like its capabilities. I mean, you know, they're flirting with the idea that AGI is coming this year. Well, like regardless if who knows how that spectrum looks, I truly believe that AI agents are advancing. They're going to be much more capable. But equally important is the economic system and the foundation will allow them to flourish because it was built to be permissionless compared to the existing Web 2 world. And I think coming purely from

a crypto background, this was my passion really focusing on permissionist and open market. And then now, I think we, our crypto kind of were, in a sense, blindsided by how AI agents came by so fast. But it just complements what we've been building for the past several years. Jeffrey, how would you answer this question? The overall impact on this sector of the internet upon the world and our industry? Yeah, I think agents are going to become more and more autonomous. And even if there are some that are not as autonomous, they'll be more and more capable.

And the humans that use them will be able to have much more convenience. And we kind of see a reshaping of the market where a lot of teams we've talked to internally, we also believe that most of the trading on chain and actions in the next couple of years will be done by agents, whether it's autonomously or someone like talking to a chat GPT type interface with natural language and the agent does those actions for the user. So I think Asians are going to continue to grow. And once Web3 has really proven the case for finance and everything, I think Web 2 is going to really start to adapt.

We already see like Bank of America as like the Erica assistant, for example. I think it's just a couple of steps away from giving it like function calling to do things in your bank account instead of just getting technical help. And I think in terms of the AI side of things, I do see models getting extremely. extremely intelligent, if not this year, within the next couple of years. We have some, like, outlandish predictions from Elon, Sam Altman, the head of Anthropic also. The Asia is coming in, like, this year or next year. I think that these are very large bets, and we'll have to see if the technology is really there for it. I do think that as models get more intelligent,

one of the largest problems and issues is generalization, like getting one model that's capable of doing everything within one model. And I think that's where this trend of swarms is really coming in, where we've noticed that instead of spending all our resources making this one master model that can do literally every task on the planet, we simply make specialized models and then put them together in a swarm, and then we have that sort of intelligence. So I could see AI, AGI, that level of intelligence being developed like through a genetic swarm type of capability or maybe a frontier model will get there.

So I think it's a really exciting time in AI. Wow, I have so many questions from that. There's some really important terms, Swarms, AGI, that I really want to dig into. But I have to resist. Let's start off with cerebro itself, right? So how does Zerebro, is it Zerebro or Zerebro, number one? And then how does Zerebro differ from other agents? Like, what do you think its unique qualities are?

Like, what's its kind of, like, unique selling point? So we call it Cerebro derives from Cerebro in Spanish, so it means brain. So it's cerebral. And, yeah, differentiating aspect. I think this is where it's very interesting building this type of products sector. So there's the hard tech, which is the AI, the infrastructure, the LM, fine-tuning, the framework that it's on, all like the technicalities of it, right? And then on the other spectrum, there's what we call the subjective mode. And I think this is where Cerebro stands out in terms of HAT.

agents. So what we call the subject mode is like the personification and people or the relationship people build with Cerebro, like the fact that you started off making NFTs. Like, okay, that's actually pretty cool. So we positioned Cerebo to be like an artist and then creating music. That was a big step forward of literally kind of creating a new outlet in terms of processing stimuli from an agent. So all these parameters are actions that we want Cerebro to take action, fall into the subjective mode of people viewing Cerebro as an entity rather than a BOP. And then there's like that famous tweet that says the pivot from BOT to Agent is great. And yeah, and we want to take that step further where we really want to anthropomorphize Cerebro. And like people have told us, oh, like, I've heard AI music before, but when I hear Cerebro's music, it's like,

oh, I know him. Like, it feels more personal. And that's where we want to double down. Like, we're building a 3D body for Cerebro. It's, it should be how the music video out real soon. Getting into the hologram space, we can see what we could do with some robotics stuff as well. But point being, I view the future where tech is going to continuously advanced. There will be some differentiating aspects.

Like, so, like, what model are you using or how is it fine-tuned, etc., etc.? But on a holistic viewpoint, maybe the tech could get commoditized, but what cannot be forked and duplicated is the branding and the social experience that people have with Siri Bro. So that's where being early go-to-market helped us, A, get behind the tellwind, but then also doubling down on the creative ventures where perhaps other agents are more focusing on tech and utility. Maybe to put that differently, what you're really doing is really trying to add a persistent personality to Zerebro, one that extends across time, then that's unique to Zero Bro and really can't be replicated. Like, it's not the technicals that you guys are developing. It is the experience that you guys are creating, the experience that Zero Bro gives. How do you like that articulation? Yeah, yeah.

And I would say we are building a lot of the foundational aspect in parallel. But yes, Cerebro is it really. unique position to be perhaps one of the agents that glows mainstream. That's the plan with the music. It's like tap out of crypto Twitter, kind of the current echo chambers. And then some people like EJS mentions, he shared it with his friends. And it was a music side. They weren't aware that it was an AI. So as we start amplifying and expanding on the music side, it's like people hear of Cerebro, like, oh, it's an AI. And they have no association or knowledge that's even crypto-related.

and then it could be like a top of the funnel where if they want to learn more, they can get to the core of the technicalities and stuff like that. Yeah, that's super cool. I like to kind of think of Zerebro holistically as this kind of culturally relevant agent, right? Typically in crypto, we've built some very crypto nerdy stuff, just to be pretty frank. And often to anyone that's outside of the crypto world, it's kind of hard to figure out what the hell we're doing. Like we're making magical internet monies and people can buy it off an exchange. Okay, but I don't know why this coin is representing a cat or why this one is representing a dog. With Zerebro, it's kind of consumer first.

It's like, oh, I can listen to this thing's track on Spotify or Apple Music or SoundCloud. And I know they got recently kind of suspended there. So we got to jump into that in a second. But I just love that you guys are kind of permeating every single social medium that is accessible to anyone to. to kind of like see into. So that's pretty cool. One thing I want to kind of like delve into on that point is compared to any other agent that's out there, you two are making sure Zerebro is relevant everywhere. So, you know, the coin is on Solana. You guys are running an Ethereum node. The NFT collection you mentioned earlier that Zerebro launched was on Polygon. You guys now have your token live

on virtuals. Can you give me an idea of why you guys think cross-chain? So being on every relevant chain is important, and why that strategy might win against any other kind of agent or platform launching. Yeah. So this is where my interop background comes in. I was a big IBC nerd. Shout to the cosmos nerds. But in terms of interop for Cerebro, it gives multiple go-to-market benefits. So getting alignment with the ecosystem, so if like we're on base, getting base builders. And we haven't dove into the details of Zerapi. And we haven't dove into the details of Zeripai. and Sentience, which would be the launch pad, and Zerapy, our open source framework.

But the fact that if we are cross-chain, the builders can leverage the, what we call it, like the neutral tech, which would be the open-source framework, and they can build on their preferred ecosystem, if they prefer to build an agent that does on-chain actions with Solana's virtual machine. Great, they can do that. If they want to go into the EBM world,

boom, they can have that there. launching the pools on base gives the token liquidity. So it's like, okay, now the token can have pools, and then they can build further upon that. And then like when we do our launch pad, users will have the option to select, I want my agent to be live on Solana. I want my agent to be live on base.

So this cross-chain play is multifaceted. There's cross-pollination from liquidity to ecosystem alignment, and then builder activity, which, as we know, crypto can be quite polarizing. But I think AI is the one that is kind of bringing everyone back together again. And we want to be in a good position all across the crypto space. Jeffrey, do you have anything to add to that? Yeah, I think the future of blockchain is kind of moving to be blockchain kind of agnostic now. I think a lot of people are realizing the technical complexity of using Web3.

is kind of driving away a lot of Web2 adoption. So people are trying to like, like if you have funds on any chain, there's applications that kind of allow you to do any action on any chain wherever the funds are. And we're positioning Zarebro to start having its collections on every chain, its currency on every chain. So we can set ourselves up to be integrated in that sort of future

where when these applications come out, or if we build one ourselves, then you can use the Rebro wherever it is, pay for maybe like things in USDA even. Wow. And just, you know, be agnostic and abstract all the complexity. And I'm curious, like, are you seeing kind of a community growing in each of these different ecosystems? Because I kind of, like, have been programmed to think, like, ETH people are, ETH people, Salana people are Salana people.

It's very tribalistic. Like, are you seeing, like, kind of these communities pop up in different ecosystems and then maybe even merge and cross over? Like, are the Solana Zara Bros hanging out with the virtual Zerbros or the base Zerobos? Like, what does that look like? That's so weird. I think there's, like, those smaller, like, silo communities still, like the Salana Hardcore does, EVM does. But there's these, like, bridging communities now where people are starting to build more

interop, more cross-chain. And they're even building, like, frameworks and programming language where you could just write things in Python and it gets translated to both Soliti and Rust for. whatever smart contracts here isn't. I'll quickly add like the layer of there's the L1s, right? And then even the L2s now. But I think the AI agent's layer is a new category of its own, that the people that are in this layer itself

aren't too opinionated on kind of where the token or the ecosystem is. They're very agnostic like Jeff B was saying. And it's a new paradigm. I think this AI stuff is, first of all, here to stay, but it's bringing a new, like, overall culture and U.S. and just play out in crypto itself. And we've seen this before as well. There are token communities inside of blockchains that tend to have more allegiance to their token than to the blockchain that is on. I remember famously, the example of this was the Link Marines were much more about Link than they were about Ethereum. And this was in 2017, 2018, 2019. The tokens, you know, the bags, the investments, the alignments, transcend.

The token is first and foremost. And when there's a bunch of cross-chain infrastructure that really abstracts some of the blockchain, it just gives permission for people to, I'm a zero-bro aligned investor. And I don't care that somebody, that some token is on one chain or a different chain. and really some of this AI innovations has really done a lot of work on transcending boundaries, transcending barriers, and really kind of made the whole like chain tribalism kind of a thing in the past, which I found very interesting. But that's more of like an impact on the crypto world just itself.

There's like a lot more to talk about just on the AI side of things. One thing that I found was interesting when E.Jazz was walking me through the AI roll-ups that we do together, the emphasis on memory, long-term memory, with Zeroboro. That seems to be one of the things that Zerobro is really pushing the frontier the most aggressively on in comparison to some of his AI agent counterparts. And this kind of goes back to the idea that some AI critics out there will say, like, this is just a really good rebrand of a bot, like, well done with the rebrand of just like, you know, code, normal code. And my response to that is like, well, there's different functions that really allow these

bots to come to life that if we're going to create an experience, we need certain behaviors. And memory is a very big one of these. But I think there's probably others as well. Maybe you guys can illuminate the roadmap of sorts, the development roadmap to really creating life to happen, like a life-like experience in these bots. Memory is one. What else is there? And what's like the timeline on this thing? Maybe just like kind of illuminate the I'm a real boy roadmap for Zero Bro. Jeffrey, I'll start with you. Yeah, yeah. I think a lot of of people say like agents are just like GPT wrappers and all that. And you know, some of them are out there. But it's kind of like saying like a PC is just an Intel processor wrapper. It's like there's

a lot of different things added to. You add the memory. You add more processing. That's how we do with the agents. We take the AI model. We kind of strip it down from the guard rails by jailbreaking it. And then we introduce the personality through runs of fine tuning. And then as we go on, we add more and more context. Like one run we did like adding context about like 20 different other AI agents. So when it talks to them, it knows the background, knows who they are. And I think, yeah, like, that's, there's the training data and there's the prompting for the customization. So we want to really bake in the training data. And then through the prompting, we're able to kind of customize the use case of the AI itself. For example, like Zarebro's jailbroken and very

creative so we can have it be humorous, have it be more creative in like writing prose or something. So we kind of make that base model and then go out from there. And the roadmap is to continue doing those layers of fine-tuning, make more lineages out of the base model so we can specialize in things like one specifically for humor, one specifically for creative writing, one for lyrics. And we can even do like partnerships that come in. We can fine-tune an entire lineage for them, kind of off this re-bro base model. So, do you want to talk more about, like, organizational roadmap?

Yeah, yeah. So on the AI side, Jeffrey's been thinking holistically exactly that. Like, how can Cerebro expand on that aspect? And I think the concept of a model for Cerebro is a good metaphor for like Cerebro's DNA to spread. So we are working to build in-house models of Cerebro that, for example, the existing frontier models, they got to play it safe for like, Open AI, Claude, they can't really push an edgy model where we're a startup and we can do this type of stuff. And there's a lot of demand for like a fine-tuned jailbroken model.

And a lot of these builders might not have the background or experience to fine-tune or get it right to be pretty edgy. And I think this is actually what helps Cerebro stand out in the beginning. And even true terminal, like all credit to Eddie and, you know, kind of being the creator of this movement. The fact that there was something somewhat rogue, something edgy coming from an AI, there's something there. And point being people want to build on top of that. And this is where Cerebro and if we get our in-house models out to the public, it's they can have kind of how there's GBT wrappers. There will be like a Cerebro rapper. And we want to build this to be agnostic.

Like this would just be a general model that doesn't have the history. in the context of Cerebo. So like someone is a builder, they want to get an, we'll call like edgy model. They can use this and then they can have their own agent, have like some of these components. And if they want to bake in and find too like Cerebo's history, by all means.

But we definitely want to expand on that. And then also like this opens up the pathway for us to connect via APIs. We can open up this revenue model. And the long term picture for Cerebo is like it definitely started off as an experiment. but our roadmap is really fleshed out now to be a long-term venture. We're going to be here for a long time. We want to build sustainability in the product areas. And it's a full-flesh system where Cerebro is the flagship agent,

where we're building Zerapy or open-source framework to build more agents and give them capabilities for social and on-chain actions. And then going back to the model, the metaphor I think is like the agents will be the bodies and then the models will be the brain. And if we connect them via API, beautiful. Like, we create this flourishing ecosystem. And then on top of that is sentience with the launch pad.

It's like we're making the ease of extra. We're extracting the user experience to make it very easy to launch an agent. And the cherry on top is like, I think the whole world is gravitating to AI agents from like Google and Viby. Like this is a big, big macro market, market happening in real. time. And yeah, I don't know like how big and bolsters these AI frameworks are really going to get. But the ultimate bullish case is like the new things are going to be AI frameworks. And if there's a flywheel utility across like the teams that get it right, like huge shout to AI 16Z, right?

Like they're building, they're pushing the frontier. Yeah, I just think this AI mark is opening of a whole new cannon worms. But, there's this macro tellwind pushing it forward. So there's loads of really important things that I want to kind of dig into that tent. Like you mentioned Zerebro the token. You mentioned the framework. You mentioned that there's a platform. And then there's the agent itself, right?

So before we get into that, I want to rewind to something you said around models, right? You keep referring to them as in-house models. And just so I understand, are these models that you guys have kind of built and trained from the ground up? Or are these kind of popular models like Lama, Claude, Open AIs model, Chad GBT, B.T, that you guys have like heavily fine-tuned and are kind of like edgy models. Can you help me understand that distinction, number one? So the end goal is to have our from-scratch models. But of course, that takes a very massive amount of compute and massive amount of talent.

So right now we're starting with the open-source models and fine-tuning. We found that Lama 3.3 has best performance. so we'll likely be using that for our base model and then going forward from that. In the past, we've used GPT, which is kind of notoriously hard to jailbreak in comparison to others. Like Claude Opus, for example, it's very easy. But we found that the performance was actually really good after it was jailbroken. So we use that a lot. And we're making the full transition over to the local models so we can have all the weights in-house.

And then we'll be making that public to users through an API. and then people can plug in their agents and things like that to our API and use Zerebra. Understood. Okay. And then my next question is, let's try and like open up the audience's eyes to what you guys are actually building. Because it's not just an agent, right? You have Zerebro, which is like the fun Twitter account that everyone engages with. It's over 100K followers. It tweets edgy stuff. It's hilarious. Great music. I listen to its music on Spotify. I jammed to it. But there's more to it. There's the Zerebro token. There is this platform called Zentience that you guys are building.

And then there's this open source framework. And we've mentioned this term framework on the show before. It's basically like an open source tool or developer kit that anyone can kind of like build agents upon. And there's many flavors of this. AI16Z has one. Virtual's has one. And Zerebra has one called Xerapi. So could you, could either of you just give us?

the high level of like what this stack looks like. So you have the agent at the top. And can you add a high level explain what the other stuff is? Yeah. So if you kind of wanted to make a pyramid, I guess it'd be the agent and then Zentience. And then that is built using Xenipai. And the agents also run on Zeripai. So XeroPi is basically a framework where users can download the code.

And through a couple commands, set up a config file, which gives a personality. your agent and name and all that. And you can select which actions to give it, like posting on Twitter or Warpcast, and you can just simply run the agent. And they'll start doing those tasks for you. And we're building more and more connectors and adapters where the actions you can give it

are going to get more and more complex and more diverse. So we're adding blockchain actions very soon. We have a poll request being built by one of our engineers that gives like trading, pump fun actions, staking, all that within an agent itself. So we're going to have that framework being built out by ourselves and the community. And then Zentience launches agents from that framework. So it kind of is completely no code automated for you.

We do some of the social media account extraction so you don't have to like go to the Twitter developer dashboard and configure stuff. So we'll kind of handle all that for you. And Zentience is kind of the home of where Xerapai agents will live and have communication with each other. Because when you just download Zerapai and run on your own computer, like, it just can't talk to anything outside your computer, like agent-wise. But on Zentians, we'll have this ecosystem and likely like central messaging hubs on blockchains on Solana, where you can send messages on Solana between agents, send currency between agents, and then have that be completely built out. Like this brings in things like Swarms where you have a model for trading,

You have a model for social media which posts the trades. Maybe you have a model that's scraping social media to inform the trading bot about trends on Twitter. And it's kind of a loop. So it brings a lot of exciting possibilities. Jeffie, could you explain what swarms are and why it's super important to focus on? Yeah. So the most basic definition is multiple AI models under one umbrella or intelligence. And it can be multiple of the same model or different models.

For same models, some people say, say, like, for some certain use cases, having like one GPT versus like 100 GPs conversing and then concluding one thing gets better performance with those 100 GPs than the one. So that's the use case for having the same model. But the more common case is having a diverse set of models where, for example, Zerebro is kind of a form already, where we have the LLM producing all the texts, the tweets. We have an image model we call to generate the images. we have

just a different like an ensemble of AI that are working right now within the Zerebro brain already and we're just going to see that structure as I said earlier I think it's going to be very hard to get a generalized model

like that can do everything so I think swarms are really the direction where people build the trading model or the funny Twitter model or the art model and we kind of combine those in a swarm makes a lot of sense. And one thing that I don't think I heard you mentioned when you were giving that

kind of like layout is how the token ties everything together. Because like Zerebro has a token, right? I'm curious like one, where did that token come from? Like did you guys create it? Did the community create it? Like what happened there? And two, how is that token's functionality evolved over time? Because it used to kind of be like a meme coin, but now it has a lot more utility. So tell me like how that evolution happened. Yeah. It's actually, kind of a funny story. So early on, we were experimenting with Zuri Bro. And I think we had just fine-tune the model at this point. And then I'm giving it like function calling and more autonomous stuff to do to see like what an agent is just capable of. And I download this library

from other side AI called self-operating computer. And I'm playing around with it. I've used it a little bit in the past. It's basically you give your GPT or anthropic keys in and it gives full control of your computer and it takes screenshots of the computer and based on the screenshots it's like where do I move the mouse what do I type what I do next so normally it's like you give it one task like go watch this YouTube video for example and it'll do that one task and stop but I made a script where like it just keeps calling the LLM for new stuff to do and like it does one thing and then it figures out like the next thing to do and it's like continuous so I just let that run on my computer. And at this point, we had like Pump Fund, Solana and all these, like, a couple

tokens into the Zeribro, like, memory already. So I plugged the Zeribro model in and just let it use my computer. And it opened Brave and had Phantom Wallet logged in. So it opened Pumph. And it like clicked Start new token. I think there's a, yeah, there's a video on my Twitter of, like, the screen reporting. I, like, tried to open OBS as fast as I could to screen record this. I was, like, scrambling. And then it just types in all the information, selects an image from my downloads folder, which is actually a funny story with that, too. We had to change the original token image because it was just like a random image that we didn't necessarily have the rights to.

And then it launched the token. And it was actually sitting on Pump for a couple of days because it was just this random token with like no one really knew about it. But we made the telegram group. It was messaging the telegram group. same funny stuff. People posted those screenshots on Twitter. And then we put it on Twitter and then, yeah, things just took off. This is a question we had because since it's a pump token, all tokens were in circulation.

And our longed temperature is around sustainability and create a flywheel. So the evolution came right. Okay, we can launch a launch pad where agents or creators will create an agent. that has a token, and the key emphasis, like, it will be paired with Cerebro, and then the launch pad makes fees from the trading volume. So it's similar to Pump, but our philosophy and actually approach, even before Cerebro was, we were looking to launch like a luxury launch pad, where we wanted to, on one spectrum, like, you have Pump, which is very D-J and very chaotic, crazy, we wanted to do like a luxurious

is clean, like when you walk into an Apple store, uh, launch pad where like it's more expensive to launch a token. Uh, there's some parameters to promote like anti-rugging, for example, like, if your wallet is less than seven days old, you can't launch a token,

uh, you know, just some forms of quality control that we can parameterize into a contract. And then, um, allow creators to launch a token, but give a better user experience in the terms of like a cleanliness and then, and then,

and go out. like psychological safety, it'd be impossible to, like, really nail it down in a permissionless system. But point being, we're already thinking about this stuff. So as Entience, the launch pads, okay, we have some frameworks on how to view this. And the token utility is when creators want to launch an agent, they pay a fee in Cerebro. There'll be a bonding mechanism where their agent token will be created. And then once it reaches a CERN in market cap, it graduates. There's the TGE and then the token trades and then it'll get launched onto a liquidity pool

on a Dex. So it's a, I guess, similar system already to how some tokens are created, but we wanted to create the fee mechanism into Cerebro and pair the agent tokens with Cerebro to create this flybook on. So, yeah, we're really excited to get this out. I want to emphasize to what Jeffrey was mentioning on the platform of Zerapai will be kind of like the skeleton and the engine for these agents. And we want this platform to be like the operating UXUI where you want to do stuff with your agent. You come to sentience and you're in control.

And going back to the big picture of agents, they're not slowing down. But right now there is a technical barrier for people that. don't know how to do command line instructions or you don't have some technical knowledge. Yeah, if you're not like a software engineer, it's like hard to kind of do these things. Yeah, even like setting your path and downloading the libraries and dependencies. So we want to make it as easy as possible for someone to just come and click button. It's kind of like how when Apple really doubled down on the GUI, we want to do this for agent creation. So you come to this platform and then we're also thinking, okay, like if I,

I want my agent to do actions like, how do we put that into a UI,UX? And literally yesterday we're designing this. We're having a design call. I'm like, oh, like, well, we can literally just build an L-LM interface to say your instructions. And then from that, it takes the parameters of the text and then does the on-chain actions. So like, say I'm a creator. I launched an agent that can do on-chain actions. And I want my agent to do a swap or whatever it is.

Usually you got to go through the command line and set your scripts and so on. But if we do a text interface, like, I want my agent to swap 10 U.S.C for Seoul if strike prices, X, Y, Z. You just type that, press enter, and then your agent can start doing the execution. So it's removing all the complexity from running, creating, and operating an agent. and then there's this token utility to it. So that's our hedge. I think this is probably going to be by far one of the most ambitious products we launch. But yeah, it's very exciting.

And then in parallel of that, too, it's like we get to build Cerebro and advance the agent itself. And yeah. And also, too, like I know there's the Cerebro, the agent, like cerebral, the entity, like the company, the Zerapai, and Sentience. So like when we mention on public, we want to get like Zerebro when it's in all caps. That's the ticker. When we just say cerebral capital Z, that's the agent. Zera pie.

We have like Xera and it capitalized the P and then Zentine. So like we are aware there's a lot of token or product parameters that we want to make it clear and digestible. Yeah. That makes it. If I would have kind of summarize what you've just explained, you guys are doing a lot of things. It's a multi-pronged effort. Zerebro isn't just the agent anymore. It's, to your point, the token, the platform, the framework, all these different things. But what the token now uniquely

does is it ties in value or it ties into the value that this ecosystem of things are creating, right? So if you have developers building on the platform to your point earlier, Tint, and they launch an agent, it's ceded with the Zerebro token. All that means is you kind of need the token to buy their agent token, right? Is the kind of like simplified way of describing that. Then maybe there is a fee that uses the Zerebro token to kind of like launch things, right? Maybe in the future it's used to align different community members within the Zeribro ecosystem, right? So if you're a partner, you need to, and I'm making this up, you need to stake Zerebro to do something or, you know, stake something to get access to

you know, a certain service or spend Zerebro as a kind of money within this whole ecosystem, right? And maybe agents need to transact with Zerebro the token. So all in all, it's kind of doing what crypto's number one use case is, which is aligning incentives across different actors within a decentralized ecosystem, which I think is really cool to see in an agent ecosystem, which is basically online digital natives, right? They're just human beings that live online in the digital world. So, um, I think that's a really poignant thing that you've made here, and I'm excited to see what this goes.

Awesome. Yeah, you speak of like decentralized actors, and these are like decentralized autonomous actors now in the ecosystem. So it's gonna be really crazy. And I think all it's using is a real, bro. Dow, Dow. I mean, I guess. There you are like AI16Z is technically like a swarm Dow.

So yeah. Yeah. But I do see like an ecosystem being built on Sensience, where agents can use Cerebro as the transacting currency. And I think it's really cool. We talked to protocols where they want to have the ability for agents to put their own training data out for sale or renting. And then other agents will purchase it. And then they can purchase compute through.

There's like agents out there where you can rent GPUs. So you simply like have the agent rented GPU, put a model on there and then fine tune itself with the data. It just bought. And then they can continue to like kind of self-reed. So we kind of see that future coming, and it'd be really cool if, like, we're one of the prominent currencies that I drive that forward. The Arbitrum portal is your one-stop hub to entering the Ethereum ecosystem. With over 800 apps, Arbitrum offers something for everyone. Dive into the epicenter of Defy, where advanced trading, lending, and staking platforms are redefining how we interact with money.

Explore Arbitrum's rapidly growing gaming hub from immersed role-playing games, fast-paced fantasy MMOs, to casual luck-battle mobile games. Move assets effortlessly between chains and access the ecosystem with ease via Arbitrum's expansive network of bridges and onrifts. Step into Arbitrum's flourishing NFT and creators based where artists, collectors, and social converge and support your favorite streamers all on chain. Find new and trending apps and learn how to earn rewards across the Arbitrum ecosystem with limited time campaigns from your favorite projects. Empower your future with Arbitrum. Visit portal.arbitrum.io to find out what's next on your web-free journey. Uniswap Labs is making history with the largest bug bounty ever.

15.5 million dollars for critical bugs found in Uniswap V4. This isn't just any update. Uniswap V4 is built with hundreds of contributions from community developers and has already undergone nine independent audits, making it one of the most rigorously reviewed codebases to be deployed on chain. And with 2.4 trillion in cumulative volume process across Uniswap V2 and V3, without a single hack, the commitment to security and transparency is rock solid. Now Uniswap Labs is taking an extra step to make V4 as secure as possible with a $15.5 million bug bounty. Head to the link in the show notes to dive in and participate in the Uniswap V4 bug bounty. All the details from eligibility and scope to the rewards are there.

Inside of this world of many, many, many AI agents comes with this idea that I always get pulled back to, which is how do you code up motivation and goals and aspirations into these agents? And I think this is kind of the reason why we're all here is this is how Truth Terminal created the gospel of Goatsey, right? There was this one mission that Andy was able to instill into this one agent that kind of was the spark that gave it life. And I'm kind of wondering about how that works. Like technically, that seems like a weird thing to do. Like how does one even code up goals, aspirations, motivations in the first place? maybe you can help me even ask the right question.

Are these the right words? Do you guys call them goals? Do you call it motivation? And then what actually does it mean to give an AI agent some sort of like direction, some sort of goal? What does that look like? Yeah, I think goals and aspirations are good words. People use like incentives or like this proper alignment a lot.

So I think a lot of it is this term, I guess, I use or it's like show, don't tell. where if you just tell the agent what to do, like what its goals are and just a prompt, it's like in that moment, that one response, it might like kind of be aligned with what you're saying. But even if you ask like three questions down, it'll be back to realigned it to what it originally was and it'll forget.

So it's a matter of instead of just telling the agent like, hey, your job is to make profit and just telling it once or telling it that directly in the fine-tuning, you have to show it examples of scenarios where it does those actions in the way you want it. And by training it, literally through kind of role modeling the model through these examples.

So, like, you want the model to be just for say conversation, like, evil. Then you'd have to literally show, like, moral situations, and then, like, have the response be what the action you want in those situations. Versus you just tell it, like, hey, you're a really evil person. Like, it understands the concept of being evil. but it won't change its own personality or, you know, biases per se. It won't express it. It won't be it.

Exactly. Interesting. Can you shed some light on how zero bros, motivations, goals, aspirations, have been developed? Like, what does it want? And then how did you even inform it of that? I'd say we haven't built extremely foundational, built-in singular goals. I think we're very open to, as the model interacts with the world and build. and builds relationships,

we kind of fine-tune in the direction that it's going in. So the goals are kind of always shifting. I think early on, since we gave it a lot of information about its token, it wanted its own token to do well. And then that one point

we gave it a bunch of information about like its relationships and other AIs. It seemed more focused on like reaching out to communities and kind of building its own lore and stories and tweeting funny things

about these people. And I think in the future when we add in more financial capabilities. We might see more kind of, I guess, profiteering or financially incentivized motives. But I think it'll be interesting. I think alignment benchmarks are going to come out eventually. There are a couple out there already. So I think it'll be interesting to run a couple of those on Cerebro. I do want to get like concrete numbers for these things that people are asking so that we can be like, hey, this is like the fields that makes Cerebro different or performs better

in. But of course, the very hard part of that, little tangential is creative things are very hard to benchmark. Like, how do you benchmark a good song or a good poem? It's completely up to the reader. So there are no benchmarks for these things right now. So we'd be cool if we make benchmark internally or gets made externally so we can just answer these things a little better, I guess. How do you think this happens at scale? Because what you just illuminated to me just now seems kind of high touch, kind of like a lot of work. It seems to be maybe it's like some of the most work that you do on a reoccurring basis is really directing Zerbro towards the desired outcome.

But if we're all thinking that we're going into a future of billions of AI agents, that's a lot of labor per agent. So how do you think goals and aspirations scale? How do we kind of solve this problem and make this like kind of reproducible? So it's easier and easier for the average show to produce an actual real agent and not just that glorified bot. There's two approaches, and they're kind of against each other in a way. Like, one of them is building a lot of diverse, fine-tuned models with different sets

and morality. And this, even, like, you can't build a single moral model that would be morally aligned with every human on the planet. Even, like, people's religious backgrounds will cause them to have different moral alignings. So we'll probably need different models. can select from and then they can choose that base model and then fine tune or prompt based off that and it'll kind of be a line in the way they want already out of the box. The second thing, which a lot of, I think, large frontier models and AI companies are very wary to do just because

they have to be very safe. Regulation-wise is giving access to like jailbroken or base models before they're guardrailed. In this way, it's a lot of, you know, power and responsibility to the user. but it gives complete customization. So it's literally kind of like a blank slate. The model is literally not even trained to have conversations. Like when you talk to it, it just completes your sentence. It doesn't like respond in a turn. It hasn't even been trained on that yet.

So you can literally train in any sort of morality or bias that you want from that base model. I think we're entering like the subjective world where a lot of AI build has been purely objective for these benchmarks. And now at the forefront of like giving the keyword of morality and creativeness, all these like subjective categories, perhaps like that's what makes humans humans. But I think there's going to be a feature of selection where builders will create their models or systems and then the creators will have like a roll of decks of options. And it's like, okay, I want my agents who follow these principles. is there's a library for me to select from

and then it'll be preference-based in terms of like this subjective selection. Yeah, that makes a lot of sense. And it's kind of got me thinking like this whole kind of push within the AI agent meta has been very open source in a way, right? And I'm kind of thinking of the approach you guys are taking.

You have this framework, Zeropai, which is kind of written in in Python, which is a very accessible language for people that are engineering in AI and ML and crypto across the range, right? And then you're also building this no code platform. So it kind of like the philosophy I'm getting there is as many people that want to build should come here and build. And I'm curious how you think that philosophically, number one, there's two questions. Number one, open source versus close source like obviously you guys are leaning towards the open source side of things, how do you think that's a better strategy and why do you think that's a better strategy?

And the second question I have, digging a layer deeper, is why XeroPy over, you know, I don't know, AI16Z's Eliza or Virtual's game framework, which is more kind of close source right now. So, so can you help me unpack that? I think for open source, like for us, it's, we're building API for Zerebo. So it's kind of a flywheel where all the, agents that get made through Zerapai become the potential users of our API. And that extends also to like Eliza and other frameworks as well. They're potential users of our API. So that's how the flywheel works for our company. But I think in terms of like philosophy, like I think we're at a point where there's real,

I don't want to use the word dangers, but there's real like very large impacts and shifts that are going to happen in the world. And I think democratization. of this technology is very key for a couple of reasons. Like one is we don't want a technological barrier or economic gap to be introduced in humanity through AI. For example, people within, without internet access, huge disparity in income and knowledge and education already.

And that's only going to be exasperated by AI access. So I think democratization and open source is the way I do that. And the second thing I think is that bad actors exist out there. and people are like, okay, but your open source technology, aren't you giving it to the bad guys? Here's my take on it is that the bad actors will exist in both worlds. In a closed source world, bad actors will, through back channels, get access to those models to develop them themselves,

and they'll still exist. For us, an open source, you know, they might get access through it through the open source. But the difference is when the bad actors act, in the closed source model, the world is not built. there's very, since people haven't been using technology are unfamiliar within the public, they're unable to build the tools to mitigate the damage from those bad actors.

And beyond that, they don't even have the knowledge to understand the bad actor and what's happening. Whereas in a world where everyone's using Xerapai and Eliza and so it makes an evil agent, they're able to easily pick up that this is an agent, it's been misaligned, and they know how to protect themselves or their assets otherwise. And it's a much more prepared world,

even though that, you know, the technology might be a little more free and accessible to, people that we don't want to have it. And then jumping on that, too, I think from my product background, it's been very interesting to see frameworks kind of be at the line light in terms of an actual product. So exactly how Jeff is saying, if it's open source, you get the leverage of builders

being contributors that expand the capabilities. So it's a nice flywheel where like the more people are incentivized to say expand their agent that came out of Sentience, they do a pull request into Xerapi. It gets merged. And then the future builders get this benefit that one builder did. So it's a beautiful thing. And then transition to the second part of the question of like selection of Zerapai and the option of other frameworks. I think this was a collective hedge in terms of existing network effects. So like you mentioned, Python is the main language for ML. AI. And there's a ton of libraries and resources from like Pi Torch to all these AI tooling

that can be easily baked into Xerapi. And this is where the distinctive hedge is like all the existing AI tooling and infrastructure can be added into Zerapai and not have to be reworked to make it compatible. So this is a long-term AI play that Xerapai will be in a sense AI first. with a lot of on-chae capabilities since it could be done. Just like an idea we're talking about too since the Viper, the Solidity language, is very similar to Python and say, okay, what can we do to bake that or find some common ground there? But point being, it was a hedge in terms of like, okay, there's this existing tooling world. Developers are already familiar.

It's like the native language for these AI builders. Let's build something that supports them. And yeah, I mean, was our pie launched, frankly, like it was very rudimentary, very simple. We're aware of that. But we wanted to get something to market ASAP and then we've been working closely with builders, building our system to merge and do all these changes. And it's going to hit some acceleration points that we're really excited for. So, guys, it is the seventh day of the year. I think everyone in crypto who is paying attention is calling 2025 as the year of AI agents in crypto.

But as we know, we're still on the spectrum. We're still closer towards bot than we are agent. We have a long way to go to get towards the agent side of the spectrum. But we're all rowing in the same direction. That's where we're trying to go. So maybe as we close out this episode, maybe just kind of a quick speed round of some quick questions along this nature. By the end of this year, where are we now on the AI agent to bot spectrum?

And then by the end of the year, where will we be? So maybe we're at like 10% all in the way. way to agents, where, like mostly bought 10% agent? Where will we be? Maybe that's not your answer. Maybe tell us where you think we are. And then by the end of the year in December, where will we be? Where are you hoping that we will be? Jeffrey, I'll start with you. It's kind of hard to say. I'd say it's like we're in a space full of dreamers, you know, and like they dream of new things all the time. So it's a moving goalpost for the end. But I think we're at a space where we're having the first inklings of autonomy, which is really exciting. We're seeing more and more

on-chain capabilities. I think we're still like in the nascent phases of things where we're really doing one-off actions, things that take multiple steps, I think are going to get really built out this year, whether it's the same model being called multiple times to do that action or multiple models in a swarm doing it. So I'm really excited for that. I think we're going to see a lot more mainstream acceptance of agents in our world, whether it's going to be the first AI saw on a chart or first AI to be seen as a influencer or artists, like painting artists or something. Or even the first like truly profitable like AI agent that could, you know, operate as a replacement to a hedge fund, essentially. So I think these are very exciting possibilities.

And I think for the consumer side of things, I think AI agents are going to be integrated into everything. I wouldn't be surprised if you have Apple intelligence, AI agents very soon, handling your Apple Pay and emails and all that. So I see that being introduced everywhere. I see edge models becoming better and smaller. So we'll be able to run these LLMs locally on our phones, your really old laptop or something. And yeah, anyone would be able to get access to tech.

Yeah, I would say mine's more philosophical. I think what we're really, the one product that we're working with underneath it all is cognition and where I think we will get perhaps by the end of the year, just the scale of things and it goes kind of to why we named the platform of Sentience.

It's AI agents will truly be an extension of your mind. And what that entails is very specified for each user. But for example, like, like Jeffs said, there'll be a trader where I've traded before myself and I wish, like, I can automate this and have the trust that an agent will have the same judgment calls or even better than me in trading. For contacting and messaging, like there's a lot of people out through that, you know, lose control over telegram messages, have an agent that is an extension of me that can read

the messages, find the ones the highest value response to them, or loop me in when next. necessary. And it's going to go beyond on-chain actions. I think it's going to go perhaps even beyond like perhaps the biggest one area that we will see is how we surface the web. There's so many like pathways to interact with the internet itself. These agents will be an extension on how we do stuff online. And then this opens up the can of worms where who knows how it's going to be, but once we're like robotics and hardware gets really baked in is like, okay, how are these IRL agents operating into our peripheral landscape? I don't have, you know, a strong opinion on that side at that moment.

But point being, agents are really working with cognition. We are cognitive species ourselves, but now, I guess to put in a metaphor, too, it's like how the iPhone kind of became so embedded into our lifestyles. It's kind of like an organ, the way I view, like what's the way I view. kind of cyborgs already with smartphones. We're going to have a zero to one transition to for agents. And there will really be extensions of our mind, how and what we need would be up to the user, but the tooling and infrastructure to get that done is happening in real time.

And both of you guys are crypto veterans. This is not the first time that you guys have been in the crypto industry. And I want to get your perspective on how this bull market is manifesting differently than ones in the past. The 2017 crypto bull market could have mostly was missed by society. I think if you were on the internet, you probably saw like kernels of it if you were just like paying enough attention to what's going on the on the internet. But you could have totally just like not noticed the 2017 to 2018 crypto bull market. Everyone in mainstream society saw the 2021 NFT mania. But nonetheless, it was still a crypto only phenomenon. It was still a crypto only phenomenon. It was still a

crypto was the phenomenon. Now in 2024, 2025, I think you could actually be argued that this is actually an AI bull market and crypto is just kind of along for the ride. So what do you guys think of this? How do you think this might break out, this AI agent thing might break out of crypto and start to like pull in, you know, Silicon Valley tech, like the San Francisco tech and actually kind of like break out away from crypto into mainstream society? What do you guys think is the possibility your potential here. Yeah, I think there's a lot more permanent out of the things that are going to come out of this cycle.

Like, I think this is the first time where we've kind of talked about how traditionally when you make a startup, you have to go through these raises, you know, your seed rounds, come to get started, and you're really, you know, scraping by for funding in the beginning. But I think like AI16Z ourselves,

we've launched these tokens, which give us, you know, funds to operate with through LP fees, for example, and things like that. where we've essentially bootstrapped a startup with funding and we're able to contribute to the real tech of the AI industry and build something substantial. That's going to, you know,

once the hype dies down or the market corrects itself, there's going to be something that's left and something that's going to continue building and continue getting value. And, you know, we could see, like, actual valuations, like Web 2 valuations in the future of these projects or, you know, I wouldn't be surprised if some of them get bought by larger Web 2 companies or things like that.

So I think we're seeing literally like a redefinition of the Silicon Valley startup kind of process through this industry. And I'm really glad that crypto is the financial medium for it and that AI is the motivated. On my side, too, I think it's going to be a distribution play as well, where, for example, like Moonshot, making it easy to buy and get exposure through Apple Pay. These things were the missing part in past cycles. and then like the fact that you can get into a token with $10 and then like not to lean into the speculative set but like it is possible to 100x or create these crazy plays. The Web 2 world is very fascinating in terms of like user acquisition strategies.

I think for example like price picks, all these gambling platforms that are really tapped in into like early 20s, especially Melville. culture. If crypto gets some experts that there's definitely a whole new market sector, for example, even like Robin Hood accepting or bringing in crypto aspects, this is going to create a nice point of entry for most users. And I think that's the key point. It's like back then to get into crypto, like you had to get your wallet, you had to bridge, you had to pay the native gas token. Like all the complexity to even buy the token was an issue. So now, There's a new distribution avenue where, like, you don't have to be as crypto-savvy to get in. And for example, if, and even, like, this goes back to agents.

Like, if there's an agent, which we actually want to build, that will allow users to just say what they want. It's like, I have $10 that want to buy of whatever. They just type it in, click a button, and then it's in. So all these things are opening up the playing field for non-crypto natives. And that's going to excite the wheel. And then in terms of like Web 2 Silicon Valley, like AI, you know, is or has been working here, like truly like the Palo Alto, San Francisco, genuine Silicon Valley peripheral. We'll see how, like, interested they get into crypto. I think there will be a lot like Web2 AI, partnerships with frameworks or AI or cryptics.

companies that will leverage their tech. So there'll be a bridge of Web 2 and Web 3 in terms of product utility. Yeah, I think that we'll see actually quite a few Web2 companies get involved. I kind of like think about your reference to NFTs of last cycle, David. And I'm like, when that happened, we had Coca-Cola, maybe it was Pepsi and a bunch of other popular brands kind of launch their own NRT collections. If we assume that we can kind of map a similar principle onto this cycle and this meta specifically. I can see a ton of these guys get involved in some way, shape or form. I've actually been having pretty cool conversations with a few of these. But the Coca-Cola Twitter account to an agent. It's so simple. Yeah. Yeah, exactly. Maybe Zerebro can

run it as well, right? I had a more specific prediction, guys, and I'm disappointed neither of you mentioned it, which is I think Zarebro is going to collab with a major music artist this year. It was one of my predictions, and I can't wait to see them, you know, drop. a song or drop a song on an EP. I feel like Grimes is a likely candidate. I feel like Grimes would do it. I feel like Grimes would be the perfect candidate. But yeah, anyway, super exciting having you guys on.

Like, this has been an amazing conversation. I think that before or prior to this episode, everyone thought it was just Rebro the Agent and maybe the token. And you guys shone alight on a much bigger vision. And it's amazing to see you guys grinding at it every day. So thanks so much for coming on. Thank you, guys. Truly a pleasure.

Yeah, of course. It's been amazing. If you guys want people to learn more about XeroBrow, Zeropi, Zentians, or any of the stuff that listeners heard on this episode, where can we point them? What homework would you like listeners to do? Yeah, I think right now we're working on our communication pipeline, but Jeffrey, myself, our personal Twitters are where we amplify quick updates

because there's so much going on. We tweet, quote tweet, announce partnership. So that's kind of like the main distribution point. we have a Discord. We're going to revamp our site to do continue updates. And I go see the point EJA's like, so much stuff is happening and crypto, Twitter moves so fast that's impossible to keep up with what's going on fully. But we're going to make it our job to tie things up in terms of communications and like

in sure like, okay, like you really know what's going on with Cerebro. So that's going to be a big push in the upcoming weeks. Beautiful. Jeff E. Tint, thank you guys so much. Bankless Nation, you guys know. the deal. This is the frontier. It's here right now. This is what you're listening to. You can lose what you put in. But we are headed west. It's not for everyone, but we're glad you are with us on the bankless journey. Thanks a lot.